Thailand

In Neural Machine Translation (and, more generally, conditional language modeling), the generation of a target token is influenced by two types of context: the source and the prefix of the target sequence. While many attempts to understand the internal workings of NMT models have been made, none of them explicitly evaluates relative source and target contributions to a generation decision. We argue that this relative contribution can be evaluated by adopting a variant of Layerwise Relevance Propagation (LRP). Its underlying &#39;conservation principle&#39; makes relevance propagation unique: differently from other methods, it evaluates not an abstract quantity reflecting token importance, but the proportion of each token&#39;s influence. We extend LRP to the Transformer and conduct an analysis of NMT models which explicitly evaluates the source and target relative contributions to the generation process. We analyze changes in these contributions when conditioning on different types of prefixes, when varying the training objective or the amount of training data, and during the training process. We find that models trained with more data tend to rely on source information more and to have more sharp token contributions; the training process is non-monotonic with several stages of different nature.

IJCNLP-AACL 2021

Analyzing the Source and Target Contributions to Predictions in Neural Machine Translation

token contributions

nmt analysis

**Welcome to ACL-IJCNLP 2021!**

The great event is jointly organized by the Association for Computational Linguistics (ACL) and Asian Federation of Natural Language Processing (AFNLP). 

As in previous years, the program of the conference includes a poster session, tutorials, workshops and demonstrations in addition to the main conference.


We were able to keep the registration fees similar to those charged for the virtual ACL 2020. The one fee allows attendance at the main conference and any/all tutorials and workshops. These fees would be $125 Regular Early and $175 Regular Late; $50 Student Early and $75 Student Late. Early registration closes at midnight July 11, 2021 (Eastern Daylight Time).

**Reminder:** It is ACL’s policy that at least one author of each accepted paper (including ACL Finding papers) must register for the conference.

**Reminder2:** Underline site will open closer to the event. If you already registered you will receive access detail

Registration is now open

The great event is jointly organized by the Association for Computational Linguistics (ACL) and Asian Federation of Natural Language Processing (AFNLP).

In Neural Machine Translation (and, more generally, conditional language modeling), the generation of a target token is influenced by two types of context: the source and the prefix of the target sequence. While many attempts to understand the internal workings of NMT models have been made, none of them explicitly evaluates relative source and target contributions to a generation decision. We argue that this relative contribution can be evaluated by adopting a variant of Layerwise Relevance Propagation (LRP). Its underlying 'conservation principle' makes relevance propagation unique: differently from other methods, it evaluates not an abstract quantity reflecting token importance, but the proportion of each token's influence. We extend LRP to the Transformer and conduct an analysis of NMT models which explicitly evaluates the source and target relative contributions to the generation process. We analyze changes in these contributions when conditioning on different types of prefixes, when varying the training objective or the amount of training data, and during the training process. We find that models trained with more data tend to rely on source information more and to have more sharp token contributions; the training process is non-monotonic with several stages of different nature.

poster

Deployed real-world machine learning applications are often subject to uncontrolled and even potentially malicious inputs. Those out-of-domain inputs can lead to unpredictable outputs and sometimes catastrophic safety issues. 
Prior studies on out-of-domain detection require in-domain task labels and are limited to supervised classification scenarios.
Our work tackles the problem of detecting out-of-domain samples with only unsupervised in-domain data. 
We utilize the latent representations of pre-trained transformers and propose a simple yet effective method to transform features across all layers to construct out-of-domain detectors efficiently. 
Two domain-specific fine-tuning approaches are further proposed to boost detection accuracy.
Our empirical evaluations of related methods on two datasets validate that our method greatly improves out-of-domain detection ability in a more general scenario.

Unsupervised Out-of-Domain Detection via Pre-trained Transformers

The advent of large pre-trained language models has given rise to rapid progress in the field of Natural Language Processing (NLP). While the performance of these models on standard benchmarks has scaled with size, compression techniques such as knowledge distillation have been key in making them practical. We present MATE-KD, a novel text-based adversarial training algorithm which improves the performance of knowledge distillation. MATE-KD first trains a masked language model-based generator to perturb text by maximizing the divergence between teacher and student logits. Then using knowledge distillation a student is trained on both the original and the perturbed training samples. We evaluate our algorithm, using BERT-based models, on the GLUE benchmark and demonstrate that MATE-KD outperforms competitive adversarial learning and data augmentation baselines. On the GLUE test set our 6 layer RoBERTa based model outperforms BERT-large.

MATE-KD: Masked Adversarial TExt, a Companion to Knowledge Distillation

Text simplification reduces the language complexity of professional content for accessibility purposes. End-to-end neural network models have been widely adopted to directly generate the simplified version of input text, usually functioning as a blackbox. We show that
text simplification can be decomposed into a compact pipeline of tasks to ensure the transparency and explainability of the process. The first two steps in this pipeline are often neglected: 1) to predict whether a given piece of text needs to be simplified, and 2) if yes, to identify complex parts of the text. The two tasks can be solved separately using either lexical or deep learning methods, or solved jointly. Simply applying explainable complexity prediction as a preliminary step, the out-of-sample text simplification performance of the state-of-the-art, black-box simplification models can be improved by a large margin.

Explainable Prediction of Text Complexity: The Missing Preliminaries for Text Simplification

NLP is currently dominated by language models like RoBERTa which are pretrained on billions of words. But what exact knowledge or skills do Transformer LMs learn from large-scale pretraining that they cannot learn from less data? To explore this question, we adopt five styles of evaluation: classifier probing, information-theoretic probing, unsupervised relative acceptability judgments, unsupervised language model knowledge probing, and fine-tuning on NLU tasks. We then draw learning curves that track the growth of these different measures of model ability with respect to pretraining data volume using the MiniBERTas, a group of RoBERTa models pretrained on 1M, 10M, 100M and 1B words. We find that these LMs require only about 10M to 100M words to learn to reliably encode most syntactic and semantic features we test. They need a much larger quantity of data in order to acquire enough commonsense knowledge and other skills required to master typical downstream NLU tasks. The results suggest that, while the ability to encode linguistic features is almost certainly necessary for language understanding, it is likely that other, unidentified, forms of knowledge are the major drivers of recent improvements in language understanding among large pretrained models.

When Do You Need Billions of Words of Pretraining Data?

Question answering (QA) models for reading comprehension have achieved human-level accuracy on in-distribution test sets.
However, they have been demonstrated to lack robustness to challenge sets, whose distribution is different from that of training sets.
Existing data augmentation methods mitigate this problem by simply augmenting training sets with synthetic examples sampled from the same distribution as the challenge sets.
However, these methods assume that the distribution of a challenge set is known a priori, making them less applicable to unseen challenge sets.
In this study, we focus on question-answer pair generation (QAG) to mitigate this problem.
While most existing QAG methods aim to improve the quality of synthetic examples,
we conjecture that diversity-promoting QAG can mitigate the sparsity of training sets and lead to better robustness.
We present a variational QAG model that generates multiple diverse QA pairs from a paragraph.
Our experiments show that our method can improve the accuracy of 12 challenge sets, as well as the in-distribution accuracy.

Improving the Robustness of QA Models to Challenge Sets with Variational Question-Answer Pair Generation

The quality of the annotated data directly influences in the success of supervised NLP models. However, creating annotated datasets is often time-consuming and expensive. Although the annotation tool takes an important role, we know little about how it influences annotation quality. We compare the quality of annotations for the task of chat-untangling made by non- experts annotators using two different tools. The first is SLATE, an existing command-line based tool, and the second is Parlay, a new tool we developed that integrates mouse interaction and visual links. Our experimental results indicate that, while both tools perform similarly in terms of annotation quality, Parlay offers a significantly better user experience.

Tools Impact on the Quality of Annotations for Chat Untangling

In this study, we investigate the role of the multiple layers in deep transformer models. We design a variant of ALBERT that dynamically adapts the number of layers for each token of the input. The key specificity of ALBERT is that weights are tied across layers. Therefore, the stack of encoder layers iteratively repeats the application of the same transformation function on the input. We interpret the repetition of this application as an iterative process where the token contextualized representations are progressively refined. We analyze this process at the token level during pre-training, fine-tuning, and inference. We show that tokens do not require the same amount of iterations and that difficult or crucial tokens for the task are subject to more iterations.

How Many Layers and Why? An Analysis of the Model Depth in Transformers

Curriculum learning has improved the quality of neural machine translation, where only source-side features are considered in the metrics to determine the difficulty of translation. In this study, we apply curriculum learning to paraphrase generation for the first time. Different from machine translation, paraphrase generation allows a certain level of discrepancy in semantics between source and target, which results in diverse transformations from lexical substitution to reordering of clauses. Hence, the difficulty of transformations requires considering both source and target contexts. We propose an edit distance between a paraphrased sentence pair as a difficulty metric in curriculum learning. Experiments on formality transfer using GYAFC showed that our curriculum learning with edit distance improves the quality of paraphrase generation. Additionally, the proposed method improves the quality of difficult samples, which was not possible for previous methods.

Edit Distance Based Curriculum Learning for Paraphrase Generation

The application of transformer-based contextual representations has became a de facto solution for solving complex NLP tasks. Despite their successes, such representations are arguably opaque as their latent dimensions are not directly interpretable. To alleviate this limitation of contextual representations, we devise such an algorithm where the output representation expresses human-interpretable information of each dimension. 
We achieve this by constructing a transformation matrix based on the semantic content of the embedding space and predefined semantic categories using Hellinger distance.
We evaluate our inferred representations on supersense prediction task. Our experiments reveal that the interpretable nature of transformed contextual representations makes it possible to accurately predict the supersense category of a word by simply looking for its transformed coordinate with the largest coefficient. We quantify the effects of our proposed transformation when applied over traditional dense contextual embeddings. We additionally investigate and report consistent improvements for the integration of sparse contextual word representations into our proposed algorithm.

Changing the Basis of Contextual Representations with Explicit Semantics

Despite the development of pre-trained language models (PLMs) significantly raise the performances of various Chinese natural language processing (NLP) tasks, the vocabulary (vocab) for these Chinese PLMs remains to be the one provided by Google Chinese BERT~\citep{Devlin2019BERTPO}, which is based on Chinese characters (chars). Second, the masked language model pre-training is based on a single vocab, limiting its downstream task performances. In this work, we first experimentally demonstrate that building a vocab via Chinese word segmentation (CWS) guided sub-word tokenization (SGT) can improve the performances of Chinese PLMs. Then we propose two versions of multi-vocab pre-training (MVP), Hi-MVP and AL-MVP, to improve the models' expressiveness. Experiments show that: (a) MVP training strategies improve PLMs' downstream performances, especially it can improve the PLM's performances on span-level tasks; (b) our AL-MVP outperforms the recent AMBERT \citep{Zhang2020AMBERTAP} after large-scale pre-training, and it is more robust against adversarial attacks.

Analyzing the Source and Target Contributions to Predictions in Neural Machine Translation

Downloads

Next from IJCNLP-AACL 2021

Unsupervised Out-of-Domain Detection via Pre-trained Transformers

Similar lecture

WikiSum: Coherent Summarization Dataset for Efficient Human-Evaluation